Clemson University TigerPrints All Dissertations Dissertations 12-2014 Assimilation of foreigners in Germany Jan Schiebler Clemson University, jschieb@g.clemson.edu Follow this and additional works at: https://tigerprints.clemson.edu/all_dissertations Part of the Economics Commons Recommended Citation Schiebler, Jan, "Assimilation of foreigners in Germany" (2014). All Dissertations. 1465. https://tigerprints.clemson.edu/all_dissertations/1465 This Dissertation is brought to you for free and open access by the Dissertations at TigerPrints. It has been accepted for inclusion in All Dissertations by an authorized administrator of TigerPrints. For more information, please contact kokeefe@clemson.edu.
ESSAYS ON IMMIGRANT ASSIMILATION IN GERMANY A Thesis Presented to the Graduate School of Clemson University In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy Economics by Jan Schiebler December 2014 Accepted by Dr. Robert Tamura, Committee Chair Dr. Raymond Sauer Dr. Scott Baier Dr. Miren Ivankovic
ABSTRACT The first chapter of this dissertation is using data from the German SOEP panel, and I analyze the assimilation of immigrants in terms of initial wage gap and assimilation rate. The analysis consists of a basic assimilation model, a cohort model, and a source country specific model. The source country specific model allows us to distinguish assimilation rates for different groups of immigrants. I find that despite having the highest education of all immigrants, East European immigrants have the largest wage gap. Secondly individuals immigrating from former East Germany have a larger wage gap than immigrants from Italy and Turkey. For East Germans I find little evidence of assimilation. In the second chapter of this dissertation is using data from the German SOEP panel, and I analyze the assimilation for immigrants in terms of initial wage gap and assimilation rate under self-selection. This paper extends the first paper by taking employment probabilities into account during the estimation process. I find that initial wage gaps in general are larger but also relative orderings between different countries of origin are different. A negative and significant lambda leads me to believe that a self-selection problem was present and was corrected by a Heckman self-selection model applied in the analysis section of this paper. ii
DEDICATION I dedicate this dissertation to my grandmother Traute Schiebler, who was unable to finish her own dissertation due to the war and my grandmother Ingrid Konopacki-Konopath who has been a great spiritual inspiration throughout my whole life. iii
ACKNOWLEDGEMENT I would like to thank my advisor Dr. Robert Tamura for his invaluable help and support. I would also like to thank the members of the International workshop who greatly contributed to my dissertation. Also, I am very grateful to my parents for their constant support and encouragement. Special thanks to Miren Ivankovic for his guidance and help. Many thanks to my classmates and friends who helped me on my journey, especially Kashema Rock, Anne Anders, Martin Richenhagen, Michael Lucas, Yannik Maden and Domenique Maden. iv
TABLE OF CONTENTS Page TITLE PAGE... i ABSTRACT... ii DEDICATION... iii ACKNOWLEDGEMENT... iv LIST OF TABLES... vii CHAPTER1 ASSIMILATION OF FOREIGNERS IN GERMANY... 1 1.1. Introduction... 1 1.2. Literature Review... 2 1.3. Data... 5 1.4. Conceptual framework and estimation... 8 1.4.1. The basic model... 8 1.4.2. Cohort model... 10 1.4.3. Source country specific assimilation model... 12 1.5. Results... 17 1.5.1. Results of the basic assimilation model... 18 1.5.2. Results of the cohort model... 19 1.5.3. Results of the source country specific assimilation model... 20 1.6. Conclusion... 23 v
Table of Contents (Continued) Page CHAPTER 2 Assimilation of foreigners under self-selection... 35 2.1. Introduction... 35 2.2. Data... 37 2.3. Model... 40 2.4. Results... 42 2.4.1. Results of the source country specific assimilation model... 42 2.4.2. Results of the educational assimilation model... 44 2.5. Conclusion... 45 vi
LIST OF TABLES Table Page 1.1. Summary statistics... 29 1.2. Basic assimilation model... 30 1.3. Cohort model... 31 1.4. Education summary... 32 1.5. Source country specific assimilation model... 33 1.6. Educational assimilation model... 34 2.1. Employment by country of origin... 51 2.2. Educational attainment by country of origin... 53 2.3. Employment by country of origin... 53 2.4. Average Employment levels by educational group... 54 2.5. Source country specific assimilation model... 55 2.6. Educational assimilation model... 56 vii
CHAPTER 1 Assimilation of foreigners in Germany 1.1 Introduction When immigrants decide to migrate into a new country they often lack knowledge specific to the labor market in that new country. They start out with a wage disadvantage, but as they live in their new host country they acquire skills necessary to succeed in their new host country and their wage adapts to that of natives. This process is called assimilation by economists and has been the center of discussion when it comes to immigration policy. As economists one of the main subjects we are concerned with are incentives, in this case the incentive for a particular individual to decide which country to migrate to. Kahanec and Zimmermann (2010) point out that the immigration of highly skilled labor and their proper management and incentivisation is of key importance for Europe. Chiswick and Miller (2009) argue in a similar direction for the U.S. and also find that high skilled immigrants are not very well matched to high skilled jobs leading them to lower income and efficiency of the labor market. In this paper three different model will be applied to better understand assimilation in Germany. The first model is the basic assimilation model that is based on the work of Chiswick (1978) and has found a lot of attention in the assimilation literature. The second model is a cohort model where the different performance of different entrance cohorts are considered that is based on the 1
work of Borjas (1985). The last model is a source country specific model where immigrants from each source country have their own assimilation rate and own entry wage gap. The model for this part of the analysis is based on an extension of the Ben-Porath (1967) model of human capital accumulation. 1.2. Literature Review The literature of economic assimilation of foreigners started with a paper by Chiswick (1978) in which he finds that immigrants start out with a significantly lower income compared to natives but catch up quickly, and even exceed natives after 10 to 15 years of residence. For this paper the 1970 and 1980 cross sections of U.S. public use samples were used. Research immediately following the work by Chiswick [Carliner(1980), DeFreitas(1980), Long(1980) and others] confirmed his findings by using essentially the same model and estimation method. In Chiswick s model the years since migration variable was considered the assimilation rate and the explanation for the high initial wage gap and the fast wage catch-up afterwards was that immigrants lack host country specific human capital such as language, knowledge of business practices, and conventions etc. Each year an immigrant spends in the host country, he acquires country specific skills and thus his income increases. Of course the wage of natives also increases over time, as they gain more experience, but the wage of immigrants grows faster, captured by the years since migration variable. Some papers also 2
argue that only a selected group of people decide to immigrate into a new country, and that this group is more ambitious and works harder compared even to the average person in their new host country. [Carliner(1980), Topel (1991)] Borjas (1985) is the first to extend the basic model by Chiswick, and he introduces the idea of cohort effects which mean that immigrants immigrating in different time periods have different quality levels and thus different assimilation rates and entry wage gaps. His analysis also led to a different view on assimilation rates themselves and also on the size of the assimilation rate and it suggested that the rates estimated by Chiswick and others were overstating the true effect. His claim was that different cohorts during different time periods were of different quality, and more recent cohorts were of lower quality compared to those entering the country before and thus had a larger wage gap. Borjas (1994) finds that the wage of successive cohorts in the U.S. continued to decline throughout the 1980 s and the entry wage in the 1980 s was 9% below that of the 1970 s and the 1990 s was an additional 6% lower. He also finds little evidence that immigrants ever reach income parity with natives. Even though the wage increases around 10% during the first two decades the starting point is too far away to converge. He concludes that immigrants stay around 15 to 20 percentage points below that of natives. He also especially points out that Mexicans and Asians seem to have an especially hard time to assimilate economically. 3
Card (2005) finds in his paper, that second generation immigrants assimilate especially well. This confirms earlier findings by Borjas (1993) and others who have done research into this aspect of assimilation theory. It seems reasonable as second generation immigrants have enjoyed the same education as their domestic counter parts. When they enter the labor market, despite some possible exceptions, they already speak the language and have the same or similar starting conditions compared to natives and thus also perform very comparable. In fact, Borjas (1992) found that second generation immigrants do better than natives in the US. There is still a negative wage gap at the beginning of the career but the second generation immigrants catch up quickly and overtake domestics of the same cohort after roughly five years. Algan(2010) on the other hand did not find this for Germany, France and the UK. However the gaps were lower than those of first generation immigrants, but there still was a gap compared to natives and they never seemed to fully catch up. In his study he focuses particularly on the UK who seems to have the largest wage gap in regards to first and second generation immigrants. He also finds that France has some immigrant groups where second generation immigrants do worse than first generation immigrants. When it comes to outmigration rates, Borjas and Bratsberg (1996) find substantial variation across origin countries exists, meaning that different source countries have different remigration rates. They also find that Belgium has the highest remigration rate and, more interestingly, that the amount of 4
remigration depends on the selection that generated the immigrant flow in the first place. Further more immigrants are more likely to return to their home when it is not far away and not poor. Dustman (2000) finds that immigrants who want to return to their home country from the start are investing less into their new host countries specific human capital. And if we assume that mainly successful immigrants stay and unsuccessful ones return to their country of origin, the number of successful immigrants increases over time. This is one of the reasons why it is important to identify country of origin specific assimilation rates. Friedberg (2000) finds in her paper that for Israel, human capital does not transfer very well for immigrants that enter Israel and she also finds that the entry wage gap of immigrants is 25% in comparison to their comparably skilled counter parts. 1.3. Data The data used for this paper is from the German Socio-Economic Panel (SOEP) and covers the years 1984 to 2011. The SOEP is a German equivalent of the U.S. longitudinal survey and it is a nationally representative longitudinal survey with approximately 13,000 households and 25,000 individuals over the course of the 27 years of data. The variables I am using for my analysis are income, age, schooling, work experience, country of origin and years since migration. 5
The sample I am using for the analysis contains 154,119 individual observations of males between 25 and 65 who are fully employed and live in former West Germany, since only about 1% of all foreigners live in former East Germany. Also only fulltime employed individuals are considered, since it is uncertain for what reasons people are unemployed. If I were to include some measure of unemployment into my analysis, I would probably overestimate the effect, since a good portion of the sample are probably unemployed because they chose to be and the data does not indicate if a person is looking for a job or not. The income used is monthly real income where 2006 Euros is the base year. The data also had a variable for weekly income data but it had too many missing or unreasonable observations so I decided to use the monthly income data. This income variable is gross income, which we are also more interested in, because when looking at economic assimilation of foreigners and domestic we are mainly interested in nominal wages rather than their after tax earnings, which could be greatly influenced by capital holdings or tax write offs etc. For a later part of the analysis I also created country of origin (or region) specific dummy variable for Spanish/Portugiese, Italian, Turkish, Yugoslavian and east- and Westeuropean immigrants. Years since migration is calculated by subtracting the immigration year from the survey year. 6
The average immigrant has been in Germany for 19.44 years, and the average age at migration is 21.5 in Germany. Turkey with 27% and Italy with 16% of all immigrants are the two biggest immigrant groups in Germany. Traditionally Turkish immigrants, especially in the 1960 s and 1970 s were guest workers whose occupational choices were limited to blue collar jobs according to the contract between Germany and Turkey. The agreement ended 1974 and the Turkish guest workers were given the choice between returning to Turkey or staying in Germany without occupational limitations afterwards. Many of the Turkish guest workers decided to stay and they are now the biggest minority in Germany. According to the Statistisches Bundesamt 1 6.75 Million foreigners lived in Germany of which are 1.73 Million Turks and 535,000 Italians. Turkish immigrants seem to be represented pretty well by the data, but Italians seem to be over represented since they only account for roughly 8% of the immigrant population but make up 16% in the data. This could either be because they are over represented in the data set or because of the fact that only male, working population is considered, and it is possible that a higher Italian population percentage falls into this category. Of the immigrant population 24% have acquired German citizenship, which is also in line with the German average. The data also include a variable for being East German as of 1990. Individuals labeled as East Germans went to school in East Germany and spent their life there until the reunification. Also included is a years since moving to 1 Equivalent to U.S. Census Bureau 7
West Germany for whom it applies. This variable is used to measure the labor market exposure of East Germans in West Germany and how they assimilate. So with the dummy and the years spent in West Germany, a comparison can be made to other foreigners, and we can compare the assimilation rates and wage differential levels. Table 1 gives a summary statistic of the data. The average years since migration in the data set is 19.4 years, which is below the national average of 24 years, but that is not surprising since individuals 65 years and older are excluded from the data. The average age at migration is 21.5 years since only first generation immigrants are considered here. 24% of the immigrants have acquired the German Citizenship and about 27% of the immigrants are from Turkey; the largest immigrant population in Germany. 16% of immigrants are from Italy and also 16% from Eastern Europe. 15% are from former Yugoslavia and 8% from Portugal and Spain. 1.4. Conceptual framework and estimation 1.4.1 The basic model The first basic model goes back to Chiswick (1978) and Borjas (1985) and is mainly there to give us a basic benchmark for later models. Its predictive powers are limited but yet give us a good starting point. In this first model we want to test the hypothesis that during different times the gap between immigrants and natives is not constant and also the assimilation rate is different. 8
This could be for different reasons like macroeconomic conditions or a change in the quality of immigrants. Generally we do however expect that immigrants start with a lower wage than natives and assimilate over time. This initial negative wage gap is due to the fact that they have not yet acquired the necessary country specific human capital. This means that they have a lower starting wage but their wage increases over time as they acquire skills including language, knowledge of country specific customs, etc. Another reason for the initial wage disadvantage of recent immigrants is information asymmetry due to the fact that employers have less information on the productivity of a recent immigrant compared to a native born with similar schooling and other characteristics. It is more costly for the employer to verify the quality of the schooling and the general quality of the worker. So he takes on a larger risk due to quality uncertainty by hiring a recent immigrant since the employer has to invest into on the job training and in Germany due to restrictions on firing people. This process could be lengthy and costly for the employer in Germany. We do however expect immigrants to have positive assimilation rates. As they spend time in the country they acquire country specific skills and learn the language and the quality uncertainty diminishes over time. The model takes the form: log = 0 + 1 + 2 + (1) 9
where is the monthly wage of individual i in the host country; is a vector of socioeconomic characteristics including age, years of schooling etc.; is a dummy variable indicating if individual i is an immigrant; and gives the number of years individual i has lived in Germany and this is set to 0 if individual i is a native. Since the vector controls for age and schooling, 2 measures at what yearly rate foreigners assimilate to natives. 1.4.2 Cohort model The cohort model is an extension of the basic model introduced by Chiswick (1978). I estimate the following OLS regression: log = 0 + 1 + (2) where represents the monthly income of person i; is a vector of socio economic characteristics including age, schooling etc. is a dummy that indicates that person i belongs to a specific immigrant cohort. This first basic cohort model only looks at the level differences between immigrant cohorts at the time of entry. It does not allow for different assimilation rates. In the second part of the cohort model analysis we now allow the assimilation rates to vary across cohorts and giving us a better picture of the economic performance of different cohorts. Borjas (1995) found a decreasing cohort quality over time for the U.S. 10
His claim is that since the 1960 s when many immigrants from Cuba immigrated, following cohort were of lower quality and thus had lower assimilation rates than earlier cohorts. In this section I estimate the following OLS regression: log = 0 + 1 + 2 + (3) where represents the monthly income of person i: is a vector of socio economic characteristics including age, schooling etc. is a dummy that indicates that person i belongs to a specific immigrant cohort. is an interaction term between the different cohorts and years since migration. 2 represents the different assimilation rates of the different arriving cohorts. In most studies of cohort effects multiple cross sections are considered which pose an identification problem since it is not possible to separately identify the effects of assimilation, cohort differences and macroeconomic effects without posing some restrictions on those relationships. In the assimilation literature, when using cross sectional data the standard assumption is that macroeconomic effects impact natives and immigrants in the same way. With this restriction macroeconomic effects have the same impact on natives and immigrants and thus assimilation and cohort effects are fully identified. It appears as if panel data has an advantage over cross sectional data in this matter, but the problem with panel data is that it could possibly introduce a time effect bias. That means it is impossible in a panel dataset to distinguish between assimilation, cohort and 11
time effects which leads us to the identifying assumption that time effects are the same for immigrants and natives. However, as Lalonde and Topel (1990) point out, this assumption of same time effects for natives and immigrants can be tricky. In their study they find that the earnings distribution in the U.S. has widened in the 1970 s and 1980 s leading to a higher wage growth at the top and lower wage growth at the bottom of the earnings distribution. However the earnings distribution in Germany has remained constant in the observation period and 20 years prior, so it is unlikely that this restriction will be an issue in this analysis. 1.4.3 Source country specific assimilation model One of the draw backs of the cohort model is that the immigrant composition is not constant over time, and it is in fact changing. It cannot be determined if a different entry wage of, for example, the 1996 2000 arrivals and the 2001 2005 arrivals is due to macroeconomic conditions, or different skill levels or simply a different immigrant composition and thus the entry wage difference would be driven by wage premium differentials between different source countries. The model used for the analysis in this section is based on the Ben-Porath (1967) model of human capital accumulation and considerations of Borjas (2000, 2013). It is a two period lifecycle model where an immigrant arrives with K units 12
of human capital that he acquired in his source country. However, only a fraction of the workers human capital is useable in the new host countries labor market. There are two depreciation factors used here are δ and, where δ represents a depreciation factor related to the general fact of the immigrant being a foreigner like quality insecurity upon arrival, and is a depreciation factor that is specific to the country of origin of the immigrant. This factor is for example influenced by language or the schooling system in that country. So an immigrant from for example Austria is going to have a smaller because he already speaks German. It could also relate to visa related issues. A worker that comes for example from another country within the EU does not need a visa, and thus it is less complicated to hire him. δ, = 0 for natives and 0 < δ, < 1 for immigrants. After immigration, an immigrant lives for two periods. In the first period the immigrant makes the decision to invest a fraction π of his full human capital into the creation of new human capital and that increases his payoff in the second period by I x 100 percent. The marketable human capital for the immigrant is then: E = ( 1 - δ - ) K. And the present value of his income stream upon arrival is: = 1 1 + 1 1 + where is a discount factor for future earnings. The human capital production function representing the increase in human capital is given by: = ( ) 13
where α, β < 1. In contrast to Ben-Porath and taking on the adaptions of Borjas (2000), a worker uses his entire human capital stock, rather than his human capital marketable in Germany to produce human capital marketable in Germany. A lawyer for example is not able to practice his profession in Germany, but we expect him to acquire human capital faster than an immigrant with less human capital, or schooling for that matter. This also relates back to the importance of the source country specific factor which could also mean that degrees required for specific jobs are or are not accepted in the new host country. As in the Ben-Porath model, preexisting human capital is an independent input in the production of new human capital. However we do not know if immigrants with higher rates of preexisting human capital acquire additional human capital at a faster rate, or maybe at the same rate, or possibly even at a slower rate than immigrants with lower levels of preexisting human capital. We can use the model to think about these questions; it is easiest to see the implications of the model, if we rewrite the human capital production function to the following form: = 1 1 + 1 This equation relates the production of human capital useable in the host country to the proportion of investments into human capital undertaken and the existing total human capital stock. The rewritten human capital production 14
function lets us separate three different cases of human capital accumulation. The first case is when α + β = 1, in which the production of additional human capital is independent of the preexisting human capital stock. This means that all foreigners assimilate at the same rate. The second case is when α + β < 1, this means that immigrants with lower levels of initial human capital gain additional human capital at a faster rate than immigrants with high levels of starting human capital. This also means that low human capital immigrants assimilate faster than those with high levels. The last case is when α + β > 1, this means that individuals with higher levels of starting human capital gain human capital at a faster rate than those with lower levels, and thus also assimilate faster. The only thing immigrants have influence on at migration is π and they maximize their post migration earnings through the optimal human capital accumulation rate given by: = 1 ( 1 1 ) 1 1 + 1 1 If α + β > 1 more highly skilled immigrants acquire more additional human capital than lower skilled immigrants, and when α + β < 1 more skilled immigrants acquire less additional human capital than lower skilled immigrants. If α + β = 1, the additional capital accumulation is independent of the starting level of human capital. The formula has another important implication, and that is the positive relationship between I and the two depreciation factors, δ and which implies that the higher the depreciation of foreign skills the higher the optimal investment 15
will be for the immigrant, and thus the faster his assimilation. This relationship exists for both of the depreciation factors and through this all sources of wage loss post migration. The relationship between I and δ, becomes interesting for natives, since for them δ, = 0 and immigrants will invest more into additional human capital than comparative natives, which makes sense because the opportunity cost for immigrants is lower than for natives, and in fact the higher the depreciation, δ and the more will be invested by an immigrant. This implies for one that immigrants will assimilate in general, and also the higher the initial wage gap the faster the assimilation. Since in the empirical part of this paper I will be looking at different assimilation rates for different countries of origin this prediction is especially interesting. In my empirical strategy I estimate the following OLS regression log = 0 + 1 + 2 + (4) where represents the monthly income of person i: is a vector of socio economic characteristics including age, schooling etc. Source is a dummy variable for the country of origin of person i. This means that 1 would measure the different entry wages for the different countries of origin. is an interaction variable of source country multiplied by the years since migration variable. This interaction means that 2 measures the different assimilation rates for different countries of origin. For people from former East Germany the interaction dummy is setup in the same way so that 1 also measures their initial 16
wage gap and 2 measures their respective assimilation rate. This specification allows me to observe a different growth path for different countries of origin and also compare them to the growth path of former East Germans. In this regression no further restrictions are required to identify 2. The only assumption is the one that was introduced earlier that time effects are the same for natives and immigrants to estimate the level and growth effects for immigrants. The second empirical model is applied to test the model estimated by the following OLS regression: log = 0 + 1 + 2 + (5) where represents the monthly income of person i: is a vector of socio economic characteristics including age, schooling etc. Edu group is a dummy variable depending which educational group immigrant i belongs to and it is multiplied with the foreigner dummy variable. Educational group 1 is less than high school, educational group 2 is high school and/or vocational training and educational group 3 is a university degree. Through this 1 measures the initial wage gap of immigrants of the respective educational groups and 2 measures the assimilation rate of immigrants depending on their educational group. 1.5 Results 17
1.5.1 Results of the basic assimilation model The regression in the basic assimilation model is broken up into five pieces, or more specifically five different time periods in order to get a first impression of how assimilation rate and initial wage gap behave over time. The results of the regression estimated through equation (1) are summarized in Table 3 and include monthly income as the dependent variable and schooling, a foreign dummy, years since migration and other covariate control variables as independent variables. Through all the different time periods the coefficient 1 is always negative indicating that for the same levels of schooling and other controls immigrants earn less than comparable natives at the time of their arrival. This is in line with the finding of Chiswick (1978) and later papers following the same model. Going back to the original model which is the one of Chiswick is based on, the Mincer-Becker model of human capital accumulation this result was predicted by the model as immigrants lack host country specific human capital at the time of their arrival. However the model also predicts that immigrants would catch up to natives as they spent more time in their new host country and assimilate in terms of income. This is indicated through the years since migration variable that is generally seen as the assimilation rate which is positive for all the observed periods, meaning that immigrants catch up to their native counter parts over time. 18
The 1984 1990 period is the pre-unification period and the following periods are all post-unification. In the period before reunification we see an initial wage gap of roughly 9% points and an assimilation rate of.4% per year which is both comparable to what comparable with studies of the U.S. labor market find. In the 1991-1995 time period that gap widens to 16.6% and peaks at 22.9% in the 1996 2000 period. However, when the initial wage gap widens, the assimilation rate usually also increases so also in this case. In the 1991 1995 the assimilation rate increases from.4% to 1.1% effectively reducing the overtaking time from 22.5 years to 15 years. The assimilation rate contracts in the 2001 2005 period to 19.1 % and 17.3% in the 2006 2011 period. As noted by Borjas (1985,1995), both the initial wage gaps and assimilation rates are highly unstable and change dramatically from time period to time period. 1.5.2 Results of the cohort model Table 3 reports the results of the cohort specific model estimated by equation (3). As in the section before the time periods are split up to give us an idea of the performance of different entrance cohorts at different time periods. In Table 3 the wage gaps at entrance are estimated and unsurprisingly all cohorts are experiencing a negative entry wage gap compared to natives of similar age and education. Also, as expected, the wage gap between natives and immigrants narrows over time. For the 1980 1990 arrivals for example the initial wage gap is -12%. After an average of 10 years in the country that gap has contracted to - 19
9% and continues to decline to -5% in the last time period. All of the cohorts experience wage growth or stable wage gaps compared to Germans except for the 1970 1980 cohort whose differential contracts to only 1% in the 1991 1995 period before extending back to 5% in the next time period. This could mean that there is a problem in the data for this particular time period or an unusual high demand for this immigrant cohort during this time period. Apart from this, all coefficients behave the way we would expect them to. Another significant finding is that the wage gap between natives and immigrants never seems to go below 5% except for the 1960 1970 arrival cohort. Another interesting finding is that the initial wage gap widens in the observed time periods. The most drastic increase is for the 2001 2005 cohort which starts at a disadvantage of -31% compared to -22% of the 1996 2000 cohort. 1.5.3 Results of the source country specific assimilation model It is instructive to begin this section with a summary statistic of educational levels represented by Table 4. East Germans living in West Germany have the highest average schooling of the sample with 12.94 years and a median of 12 years of schooling. Second highest is West Germans with a mean of 12.56 and a median of 11.5 years, closely followed by eastern Europeans with a mean of 11.77 and a median of 11 years. The gap to all the other immigrants is quite substantial as they have a mean of around 9 years of schooling. However the 20
percentage of blue collar workers among the eastern European immigrants is roughly the same as for the other immigrants that have around 9 years of average schooling according to the Statistisches Bundesamt. I will come back to this fact and the issue this fact might create in the estimation process at a later time. For the empirical analysis I separated the immigrants into different countries of origin or tabulated them into a group of immigrants with similar characteristics. The different groups are Eastern Europe, Yugoslavia, Italy, Spain/Portugal, Turkey, East Germany, and all others are listed under other. 2 For the actual estimation process dummy variable were created for the different country groups to measure the initial wage gap at arrival of the different sending countries and an interaction term between country of origin and years since migration gives the source country specific assimilation rate. Table 5 reports the results of the source country specific assimilation model estimated through equation (4). It appears that Eastern Europe has the largest initial wage gap of all other countries. As mentioned earlier this result could be due to the fact that Eastern Europeans in the sample have a high education level but are performing similar jobs as immigrants from other countries of origin. However in the estimation process, the Eastern European immigrants are essentially compared to Germans with higher levels of education and most likely also higher paid jobs. If we see this result in light of our model, 2 The exact composition can be found in the appendix 21
immigrants from Eastern Europe have a higher and are not able to use their full human capital. On the other hand immigrants from Eastern Europe do assimilate at the highest rate which was also predicted by that model that immigrants that are not able to use their full human capital assimilate faster, because a higher human capital level means that they can probably acquire new host country specific human capital at a faster rate than immigrants with lower levels of human capital, and their opportunity cost to acquire more host country specific human capital is lower than that of immigrant who are earning more. The two countries with the smallest initial wage gaps are Turkey and Italy, which also represent the two largest immigrant populations. Zhu, Liu, Painter (2013) find in their paper that immigrants that join larger ethnic communities do better than those who do not have any of their countrymen living in their new host country. In terms of our model that means that the depreciation factor is lower for immigrants of Turkey and Italy because they possibly have smaller language adjustment period since a person coming from Turkey for example can move to Berlin and find a job without speaking German, because the Turkish speaking community there is so large. Also consistent with the model is that finding that immigrants from Italy and Turkey have the lowest assimilation rates of all immigrants. Another interesting finding is that East Germans have a higher initial earnings gap than Turkish, Italian and even immigrants from Yugoslavia. I was 22
also not able to measure an assimilation rate for them, as the assimilation rate for East Germans in insignificant. This is especially surprising since East Germans already speak German upon arrival. The second column of Table 5 reports the results of equation (4) with the addition of experience as another control variable. The results change slightly, East Germans now have a positive and significant assimilation rate and some of the initial wage gaps change slightly but generally the results stay very comparable. Table 6 reports the results of the educational specific assimilation model estimated through equation (5). The results indicate that immigrants of educational group 3 have the highest initial wage gap. Relating this result back to the original model means that high education immigrants have the highest depreciation factor and thus their income is penalized the most. Immigrants of educational group 2 perform better than immigrants from educational group 3 but worse than immigrants from educational group 1. It seems that the higher the education the higher the initial wage gap to comparable natives. However the higher the educational level the higher is the assimilation rate. In terms of the model this means that α + β > 1 and individuals with higher starting levels of human capital acquire additional human capital faster. 1.6. Conclusion 23
This paper discusses a broad spectrum of assimilation analysis for immigrants in Germany and presents a series of results. The data used for this paper is drawn from the rich panel dataset of the SOEP and includes the years 1984 2011. The analysis includes the basic assimilation model where the findings in the existing literature are confirmed that immigrants experience a rather large initial wage gap and close that gap over time. As has also been documented before the initial wage gap and the assimilation rate are unstable overtime. In the cohort model we find sizeable cohort effects with the entry wage gap widening for later cohorts but the wage gap closing as time is spent in the new host country. However, this change in the initial wage gap may very well be cause by a change in the composition of the immigrant cohorts, and in fact, the country of origin composition is changing over time in the data set. This leads me to the adaption of the source country specific assimilation model where each country of origin has its own initial wage gap and own assimilation rate. One of the biggest and most surprising findings was that people migrating from former East Germany in fact have a larger wage gap than immigrants coming from Turkey and Italy. The predictions of the model presented for the source country specific assimilation have been fulfilled as immigrants that start at a greater wage disadvantage assimilate faster than those with a smaller initial wage gap and immigrants from different countries of origin have different wage gaps leading us 24
to believe that the depreciation factors for human capital in the new host country do in fact depend on the source country. The finding that immigrants from Eastern Europe which have high values of human capital start at a much greater wage disadvantage than immigrants with low values of human capital is however troubling since Germany relies on the influx of high skill labor. Germany in particular, but Europe in general, both have more restrictive laws when it comes to hiring new labor such as a higher minimum wage which makes Germany especially attractive for immigrants with low levels of human capital. 3 However, in this social market economy the upwards earning potential is lower than in for example the U.S. which means that high human capital immigrants are more likely to immigrate into the U.S. Roy (1954) wrote in his paper that the chances of high incomes attract high skill immigrants and higher job security tends to attract low skilled immigrants. The EU has undertaken steps to make migration between its member countries easier by reducing barriers of inflow and recognizing each others qualifications. In terms of policy it is crucial to continue to reduce barriers of entry to attract more high skill labor. The necessity for these reforms can especially be seen in the educational group specific assimilation model where immigrants with the highest levels of human capital have the largest entry wage gap which is troubling when a country tries to attract high human capital immigrants. 3 For comparison, according to the U.S. census bureau 30% of immigrants immigrating into the U.S. belong to educational group 3, compared to roughly 10% for Germany. 25
References Antecol, Heather, Peter J. Kuhn, and Stephen J. Trejo. 2006. Assimilation via prices orquantities? Sources of immigrant earnings growth in Australia, Canada, and the United States. Journal of Human Resources 41: 821-840. Aydemir, Abdurrahman, and Mikal Skuterud. 2005. Explaining the deteriorating entry earnings of Canada s immigrant cohorts, 1966 2000. Canadian Journal of Economics 38: 641-672. Baker, Michael, and Dwayne Benjamin. 1994. The performance of immigrants in the Canadian labor market. Journal of Labor Economics 12: 369 405. Barth, Erling, Bernt Bratsberg and Oddbjørn Raaum. 2004. Identifying earnings assimilation of immigrants under changing macroeconomic conditions. Scandinavian Journal of Economics 106: 1-22. Ben-Porath, Yoram. 1967. The production of human capital and the life cycle of earnings. Journal of Political Economy 75: 352-365. Borjas, George J. 1985. Assimilation, changes in cohort quality, and the earnings of immigrants. Journal of Labor Economics 3: 463-489. Borjas, George J. 1992. National Origin and the Skills of Immigrants in the Postwar Period. In: George J. Borjas and Richard B. Freeman, eds., Immigration and the work force: Economic consequences for the United States and source areas, 17-47. Chicago: University of Chicago Press. Borjas, George J. 1995. Assimilation and changes in cohort quality revisited: What happened to immigrant earnings in the 1980s? Journal of Labor Economics 13: 201-245. Borjas, G. J. (2013). The slowdown in the economic assimilation of immigrants: Aging and cohort effects revisited again. National Bureau of Economic Research Working Paper 19116. Card, D. (2001). Immigrant inflows, native outflows, and the local labor market impacts of higher immigration. Journal of Labor Economics19,22-64. 26
Card, D. (2005). Is the new immigration really so bad? Economic Journal115, F300- F323. Card, D. (2009). Immigration and inequality. American Economic Review 99, 1-21. Chiswick, Barry R. 1978. The effect of Americanization on the earnings of foreign-born men. Journal of Political Economy 86: 897-921. 46 Chiswick, Barry R. 1986. Is the new immigration less skilled than the old? Journal of Labor Economics 4:168-192. Chiswick, Barry R., and Paul W. Miller. 1999. English language fluency among immigrants in the United States. Research in labor economics 17: 151-200. Douglas, Paul H. 1919. Is the new immigration more unskilled than the old? Journal of the American Statistical Association 16: 393-403. Duleep, Harriet Orcutt, and Daniel J. Dowhan 2002. Insights from longitudinal data on the earnings growth of U.S. foreign-born men. Demography 39: 485-506. Duleep, Harriet Orcutt, and Mark C. Regets. 1997. Immigrant entry earnings and human capital growth: Evidence from the 1960 1980 censuses. Research in Labor Economics 16: 297-317. Edin, Per Anders, Robert J. LaLonde, and Olof Åslund. 2000. Emigration of immigrants and measures of immigrant assimilation: evidence from Sweden. Swedish Economic Policy Review 7: 163 204. Glenn, Norval D. 1976. Cohort analysts futile quest: Statistical attempts to separate age, period, and cohort effects. American Sociological Review 41: 900-904. Green, David A. 1999. Immigrant occupational attainment: Assimilation and mobility over time. Journal of Labor Economics 17: 49-79. Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2009. Cultural biases in economic exchange. Quarterly Journal of Economics 124: 1095 1131. 27
Katz, Lawrence F., and Kevin M. Murphy. 1992. Changes in the wage structure, 1963-87: Supply and demand factors. Quarterly Journal of Economics 107: 35-78. LaLonde, Robert J., and Robert H. Topel. 1992. The assimilation of immigrants in the U.S. labor market. In: George J. Borjas and Richard B. Freeman, eds., Immigration and the work force: Economic consequences for the United States and source areas, 67-92. Chicago: University of Chicago Press. Lazear, Edward P. 1999. Culture and language. Journal of Political Economy 107: S95- S126. Lemieux, Thomas. 2006. Increasing residual wage inequality: Composition effects, noisy data, or rising demand for skill? American Economic Review 96: 461-498. Lubotsky, Darren. 2007. Chutes or ladders? A longitudinal analysis of immigrant earnings. Journal of Political Economy 115: 820-867. Lubotsky, Darren. 2011. The effect of changes in the U.S. wage structure on recent immigrants earnings. Review of Economics and Statistics 93: 59-71. Schoeni, Robert F. 1997b. New evidence on the economic progress of foreign-born men in the 1970s and 1980s. Journal of Human Resources 32: 683 740. 28
Table 1.1. Summary statistics Years since migration 19.44 Age at migration 21.5 German Citenzenship 0.24 Country of origin Turkey 0.27 Italy 0.16 Eastern Europe 0.16 Former Yugoslavia 0.15 Other Western 0.04 Asia 0.05 Spain / Portugal 0.08 29
Table 1.2. Basic assimilation model Variable 1984-1990 1991-1995 1996-2000 2001-2005 2006-2011 Schooling 0.076 0.072 0.069 0.080 0.0822 (Years of education) (0.0011) (0.0013) (0.0012) (0.0010) (0.0010) Foreign Dummy -0.0956-0.166-0.229-0.191-0.173 (0.019) (0.0217) (0.022) (0.025) (0.032) Years in Germany 0.0047 0.011 0.011 0.0055 0.0041 (0.0013) (0.0024) (0.0025) (0.0025) (0.0013) - - -0.000011-0.00026-0.00019 Years in Germany squared 0.0000215 0.0000307 (0.0000722) (0.00006) (0.00006) (0.00005) (0.00005) Notes: Standard errors are reported in parantheses 30
Table 1.3. Cohort model 1984-1990 1991-1995 1996-2000 2001-2005 2006-2011 cohorts 2001-2005 arrivals --- --- --- --- -0.317 (0.0509) 1996-2000 arrivals --- --- --- -0.222-0.191 (0.0339) (0.0335) 1991-1995 arrivals --- --- -0.24-0.166-0.140 (0.0173) (0.0170) (0.0181) 1980-1990 arrivals -0.128-0.0906-0.077-0.0669-0.0589 (0.0211) (0.0179) (0.0159) (0.0166) (0.0177) 1970-1980 arrivals -0.074-0.0146-0.0530-0.0524-0.0752 (0.008) (0.0102) (0.0129) (0.0144) (0.0178) 1960-1970 arrivals -0.0374-0.0400-0.0562 --- --- (0.0086) (0.0127) (0.0179) Notes: Standard errors are reported in parantheses 31
Table 1.4. Education summary Educational levels by country of origin Mean 25 percentile 75 percentile Median (In years of education) West Germans 12.56 10.5 14.5 11.5 Eastern Europe 11.77 10.5 12 11 Yugoslavia 9.88 9 11 10.5 Italy 9.29 7 10.5 9 Spain 9.44 9 11 9 Turkey 9.7 9 10.5 9 Other 11.15 9 13 10.5 East Germans in West Germany 12.94 11.5 14.5 12 32
Table 1.5. Source country specific assimilation model Initial wage gaps by country of origin (1) (2) Eastern Europe -0.24-0.24 (0.012) (0.012) Yugoslavia -0.115-0.14 (0.02) (0.02) Italy -0.055-0.07 (0.019) (0.018) Spain -0.132-0.15 (0.03) (0.03) Turkey -0.083-0.09 (0.0148) (0.014) Other -0.112-0.11 (0.0143) (0.014) Eastern Germans -0.128-0.17 (0.019) (0.019) Assimilation rate by country of origin Eastern Europe 0.0055 0.006 (0.0007) (0.00070) Yugoslavia 0.0042 0.0019 (0.00093) (0.00092) Italy 0.0021 0.0010 (0.00083) (0.0008) Spain 0.0043 0.0053 (0.0013) (0.0012) Turkey 0.0028 0.003 (0.0007) (0.00072) Other 0.0023 0.002 (0.00061) (0.00060) Eastern Germans -0.0004 0.0026 (0.001) (0.0013) Notes: Standard errors are reported in parantheses 33
Table 1.6. Educational assimilation model Initial wage gaps by educational group (1) (2) Education group 1-0.059-0.053 (0.011) (0.011) Education group 2-0.181-0.173 (0.009) (0.0092) Education group 3-0.236-0.229 (0.018) (0.018) Assimilation rate by educational group Education group 1 0.0018 0.0017 (0.00053) (0.00053) Education group 2 0.0045 0.044 (0.00042) (0.00042) Education group 3 0.0065 0.0066 (0.00084) (0.00084) Notes: Standard errors are reported in parantheses 34
CHAPTER 2 Assimilation of foreigners under self-selection 2.1.Introduction The estimation of labor market performance of immigrants once they enter their new host country has been performed by many scholars starting with the human capital theory of Mincer (1974) that incorporated schooling in a regression analysis. The human capital theory was then transformed by Chiswick (1978) and Borjas (1985) to estimate and explain the labor market performance of immigrants. In the early papers by Chiswick and Borjas immigrants assimilated rather quickly in their new environment and some concern was raised about the estimation methods as cross section data was used that did not allow it to follow an immigrant over a time span and that was only available every ten years. The cross section could give biased results if remigration is not taken into consideration or, as pointed out by Borjas, a quality change in the immigrants is observed and not corrected for. Borjas (1985) has argued that the decline in cohort quality was responsible for a biased increase in estimated assimilation rates for immigrants. This interpretation has however been disputed by Duleep (1996), Regets (1997), and other researchers. Most economic papers written about economic assimilation use an OLS approach in their empirical model where wage is regressed on personal covariates and a years since migration variable that is capturing the economic 35